- Design
- Prospective cohort with unsupervised clustering; replication in a national registry
- Population
- 600 adults aged 18–69 with recent-onset type 1 or GAD-positive diabetes (German Diabetes Study); 483 in the DPV registry
- Primary outcome
- Subtype structure, beta-cell function trajectory and complications
- Effect
- Mixed subtype about 6.1 years (95% CI 4.6–7.8) earlier on beta-cell decline; retinopathy HR 4.93 (1.71–14.24) vs classic
A prospective German cohort of 600 adults diagnosed with type 1 diabetes within the past year, together with GAD-antibody-positive people first labelled as type 2, used clustering on GAD antibody status, fasting C-peptide, glucose, triglycerides and HDL cholesterol. Mean follow-up was about six years, and the structure was reproduced in a separate registry of 483 adults. It appeared in Lancet Diabetes & Endocrinology on 30 October.
Three groups emerged. Classic type 1 was lean, insulin-sensitive and had poor beta-cell function. A mixed group was antibody-positive but heavier and insulin-resistant, with better beta-cell function. A third group was antibody-negative with lower genetic type 1 risk. The two antibody-positive groups had the same autoimmune and genetic profile; modelling placed the mixed group about 6.1 years earlier along the same decline. The mixed group had more retinopathy than the classic group.
The practical point is that an insulin-resistant adult with GAD antibodies is not a different disease, just an earlier point in the same one, and will probably lose more beta-cell function with time. The cohort is observational and the subtypes are statistical groupings, not diagnostic categories.
- An overweight, insulin-resistant adult with GAD antibodies may be on the same path as classic type 1, only earlier.
- Plan for falling C-peptide over time in antibody-positive adults, even if they start on oral agents.
- Screen for retinopathy promptly in antibody-positive adults who also have metabolic syndrome features.
- Antibody-negative 'type 1' carries lower genetic type 1 risk; review the diagnosis when the course is atypical.
Why it matters
It reframes so-called latent autoimmune diabetes as an earlier stage of type 1, which changes how closely those patients are followed.
Don't overread it
These are data-driven clusters in a German cohort; they are not validated diagnostic categories and do not yet guide therapy.
The statistics, in plain English
The retinopathy hazard ratio of 4.93 (95% CI 1.71 to 14.24) for mixed versus classic type 1 is large but imprecise: the true figure could be anything from under twice to fourteen times the risk. The '6.1 years earlier' is a modelled position on a decline curve, not directly observed time.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for diabetes & endocrinology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free